skills/gandazgul/runwield/guide/Gen Agent Trust Hub

guide

Pass

Audited by Gen Agent Trust Hub on Sep 18, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted data from external sources and the repository itself while possessing execution capabilities.
  • Ingestion points: Processes content from external websites (web search and page fetching) and local repository files including documentation, source code, and Git history (SKILL.md).
  • Boundary markers: Instructions emphasize verifying facts against durable artifacts rather than relying on memory or subagent findings alone. It explicitly states to distinguish intent from current behavior.
  • Capability inventory: Can execute discovery-only shell commands (e.g., git log, git show, ls, grep), perform web searches, delegate tasks to subagents, and create/edit .md files.
  • Sanitization: No specific sanitization methods are described for handling external content before it is processed by the agent or included in documentation outputs.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 18, 2026, 12:01 AM
Security Audit — agent-trust-hub — guide